Afrodite-Ohki opened this issue on Mar 31, 2023 ยท 2000 posts
shvrdavid posted Mon, 06 November 2023 at 11:16 AM
Wattage overhead of a power supply has basically nothing to do with dirty power and crashes. Any power supply can start to create dirty power when the system loads change quickly. The screen going black under heavy load changes is the primary sign of this. If the GPU driver fails in Windows, and it doesn't matter what driver it is, the screen still displays. It may change resolution and flash, but it doesn't go black from a driver failure. The GPU driver is a second layer, and the windows driver, is the primary driver.
If there is dirty power or worse yet an under volt during the load change, the screen goes black and programs crash. Even if you have a 1400+ watt supply this can happen during a transient power change or spike.
Wattage does not equal stability, and it never has. Consistent clean power is the key.
The 3000 series cards are the worst offenders in the Nvidia line up. A 3000 series gpu can pull from a normal of 200-350 watt under load, and spike well over 600 for a few milliseconds. If the power supply can not provide clean power during that transient spike, or worse yet there is a voltage sag, the screen will go black, and programs crash. Oddly enough, the studio drivers try to prevent those spikes from occurring by clock and wait state changes. Ironic that the driver masks the problem, that is still going to be there....
I have seen this many times, and know many people that were having the same issues that they did not think was a power supply problem because they had "enough wattage". Those systems had 850 to 1400 watt supplies that could not handle transient changes in power output, and crashed when they occurred. When the power supply was replaced the problem magically went away....
All GPU's and CPU's that have very quick load changes can push any power supply past it limits of providing clean power. It doesn't matter how many watts of overhead the supply has either.
In the end, the only difference is if the power supply makes dirty power or has a voltage sag during the outrush when the spike occurs.... If it is under voltage or sagging, it will eventually destroy most of the system components. Some slowly, some just die out of the blue, like hard drives....... When the voltage goes down or sags, the amperage goes way up.... Amps are what lets that magic smoke out of electronics, and you can't put it back afterwards.....
Power supplies are one of the cheapest parts in a computer. And also the only one that can kill the entire system if let go after it has already shown it is going bad........
The screen going black is a sure sign of that.....
Just to give an odd example. The system I am typing on is a 13th gen Intel system, 64 gig of memory, 22+ terabytes of storage, two RTX Nvidia gpus. The system has a server grade 750 watt power supply. Yes, 750, and that is not a typo, it can handle 1000+ spikes. Technically it is too small of a power supply, and it never misses a beat with the system fully loaded..... There are no voltage sags, no dirty power, at all.... I know that, because I tested the supply to 850 watts...
Poser rarely crashes on this system, neither do any other programs. Wattage does not equal stability, and it never has. I use the 4000 series gaming drivers..... And one of the cards is not a 4000 series card....
The core of the drivers, is basically the same on all RTX drivers..... With the acceptation of stability fixes like preventing your power supply from going belly up, in the studio drivers.....
Some things are easy to explain, other things are not........ <- Store -> <-Freebies->